The Space Race of the 1960s spawned a popular myth: NASA spent millions developing zero-gravity pens while the Soviet Union simply used pencils. Though untrue (both used American-made pens), this tale perfectly captured how resource constraints can spark innovative solutions. Today, a similar David-versus-Goliath story is unfolding in the global AI race.

DeepSeek, a little-known Chinese company from Hangzhou, has disrupted the AI landscape dominated by tech giants. In December, it released its traditional V3 large language model, followed by the R1 reasoning model in January. These models achieved performance levels comparable to those from OpenAI, Anthropic, and Google—but at a fraction of the cost.

What is remarkable here is how DeepSeek's engineers optimised their limited resources. They developed innovative software to train their models instead of copying expensive ways adopted by Western companies, and faced with US restrictions on advanced Nvidia AI chips, they maximised efficiency with available hardware. More significantly, they made their open-source models accessible to anyone with a computer and internet connection, without asking for subscription fees like others.

This breakthrough rattled US markets as investors who were backing big spenders like OpenAI, Google and Meta and high end chip maker Nvidia – counting on their unrestrained growth – suddenly saw red flags ahead. This naturally put DeepSeek under the global spotlight and many analysts raised concerns about their censorship and data handling.

Such........

© Mathrubhumi English